8 May, 2019
Many simulation packages claim they can help you maximize performance while also saving on energy costs, but how exactly do they do it?
This blog outlines how to calculate energy savings in 6SigmaRoom, the industry’s leading digital twin software for data centers.
This blog will:
1) Briefly describe how the 6SigmaRoom user can compile information from five primary categories (Power, Airflow, Chilled Water [CW] System, Power Usage Effectiveness [PUE], and IT) to assess data center energy impact.
2) Explain the digital twin process of directly impacting energy savings.
Power objects, such as UPSs and PDUs, operate more efficiently when running at a higher load. 6SigmaRoom can analyze how to push more load through power objects to operate more efficiently.
As such, you can effectively capacity plan to load up PDUs in stages—loading each as much as possible before bringing another online—rather than having a large number of PDUs at low load.
By increasing the efficiency of PDUs and transformers, you can minimize the contribution of electrical equipment to your data center cooling load, resulting in operating a data center at the lowest cost possible.
Fans have some mechanical inefficiency that adds heat to the airflow. This follows the fan affinity laws in that, as the user reduces the revolutions per minute, they also reduce the power needed to drive the fans and the heat added to the air stream.
The fan efficiency calculations can be completed and different solutions tested in 6SigmaRoom by, for example, running simulations at lower fan speed control setpoints or reducing the number of perforated tiles in the data hall. You can also make sure that all IT/cabinets remain within their thermal operating limits while supplying less airflow.
We are always working to improve 6SigmaRoom to suit our users’ needs. We have heard your feedback and are incorporating improvements to the 1-D network in Release 14. In R14, 6SigmaRoom will be capable of calculating individual CRAH performance based on the CW supply lines. Increasing the CW supply temperature improves Chiller performance, meaning more free cooling hours, reduced CW power requirements and lower system costs.
Our software factors in a supply temperature-dependent Coefficient of Performance (COP) of the CW system. COP can be directly translated to kW/Ton, which is the kW electrical load required to run the CW system for every Ton of cooling load it provides. This allows for estimation of the CW system power, based on the CW supply setpoint and how much cooling load is removed from the data center by the CRAHs. You can expect to reduce CW system power by 1-2% for every 1°F that the CW supply temperature is increased.
6SigmaRoom calculates a PUE value in the Room Summary. This value is quantified from a compilation of IT equipment, Cooling units and CW system powers, along with lighting loads and UPS/ PDU losses. Our software demonstrates how much PUE can be improved by raising temperature setpoints in the data center while staying within the thermal safety margins required to adhere to ASHRAE regulations.
IT airflow rates will respond to inlet temperatures, meaning that an increase in data center temperatures will also increase energy costs as the IT fans ramp up. 6SigmaRoom allows you to predict the increase in airflow from servers due to increasing air temperature (i.e. higher air temperature results in server fans running faster to cool the electronics).
To do this, the engineer will need server-level data on volumetric flowrate output versus inlet air temperature provided by the manufacturer. If this data is not readily available, the user can use the default ASHRAE or ENERGYSTAR characteristic flow curves already in the software. For more information regarding IT flow curves, please refer to the following blog: New EnergyStar Option for Defining Airflow.
How to Directly Influence Energy Savings with a Digital Twin
The typical process starts with building a digital twin of the actual facility in 6SigmaRoom. The physics-based digital twin can predict the inlet temperature at the inlet of each IT equipment in the data center.
These predictions allow the user to establish a proper baseline and assess the current conditions of the data center. The next step is to fix any issues pertaining to the elevated inlet temperatures.
Once the hot-spots have been removed, the subsequent step is to examine how best to:
1) Reduce fan speeds
2) Turn off CRAC units
3) Raise air temperature
Options one and two are straightforward and can be evaluated in the Room Summary section of 6SigmaRoom quite easily. In some cases, it can be a combination of both these strategies listed above. Once you reach a suitable fan speed, the software will return total ACU power (fan power consumption) in the Room Summary. You can extrapolate this value out to an annual kWh value by multiplying by 8,760 hours/year. Then, simply apply the blended utility rate for the data center ($/kWh) to calculate the annual cost to run the cooling unit fans.
Figure 1. You can view the fan speed on the CRAC unit’s (right) property sheet (left)
Option three is more involved as it requires considering the cooling infrastructure that is outside the data center. For example, if your maximum IT inlet temperature is 75°F, and you are comfortable running up to 80°F, then you can assume that your system can support a 5°F CW supply temperature increase. Again, you can evaluate the inlet temperature of the IT equipment with 6SigmaRoom's in-built visualization features.
With CW energy savings, you can assume that every 1°F rise in CW supply temperature will result in a 1-2% reduction in the input power needed to run the CW system. In your model, you can specify the chiller COP curve or rating based on varying supply temperatures.
All of this data is reported in the digital twin for further evaluation. You can now change the supply temperature set-points and see the potential CW savings in the software. A higher supply temperature will yield a lower power consumption at the chiller unit.
Figure 2. A Chiller COP curve
The added benefit of raising the CW temperature is the ability of the data center to harness “free-cooling.” At a higher chiller water set point, you can potentially use air to cool the chillers to provide the necessary cooling. The percentage of free cooling hours will vary based on the external air temperature and the desired CW supply temperature.
Tools such as DOE2.com and eQUEST EnergyPlus software enable you to accurately calculate the amount of free cooling hours based on external air temperatures. This will allow you to quantify the overall savings from data center to chiller.
Save Energy Without Compromising Performance
Data center performance is a delicate balance between capacity, efficiency and compliance. Increasing efficiency by raising air temperatures without proper analysis can compromise the long-term capacity of the data center or put IT equipment at risk of failure. The use of a data center digital twin allows you to find the optimal balance to save energy and maximize data center performance.
Blog written by: Akhil Docca, Director of Marketing & Mark Fenton, Product Manager
Other Recent Posts
The Digital Twin for Hyperscale Data Centers
In a world where artificial intelligence and machine learning are the new frontiers, cloud applicat…
CFD Simulation Should Be a Data Center Best Practice
Over the years that I have been involved in the data center industry, I have read numerous articles…